SCORE is a standardized way to compare the expected return and risk of fully specified card-counting games. It stands for Standardized Comparison of Risk and Expectation and was developed by Don Schlesinger. It is not a general quality score, a prediction of profit or a substitute for a simulation.
On this page
QFIT’s explanation of SCORE describes it as win rate per 100 hands with a $10,000 bankroll, a 13.5% risk-of-ruin convention and a standard set of conditions. Those inputs are comparison assumptions, not advice to accept that risk or supply that bankroll.
Where formal SCORE applies
Traditional SCORE concerns a persistent blackjack shoe and a defined advantage-play method. In ordinary RNG games that generate each round independently, past exposed cards do not provide the remaining-composition information used by a shoe count. Live video uses physical cards, but it only supports a comparable model when the actual shoe, penetration and operating conditions are known.
The broader habit of checking rules is useful in any format. Do not call a checklist a SCORE calculation, however. A formal number requires consistent simulation assumptions.
Inputs behind a SCORE result
A published SCORE belongs to one complete model. At minimum, record the game rules, deck count, penetration, count, playing indices, bet spread, optimized ramp, whether negative counts are played or observed, table occupancy and any round-rate assumption.
1. Betting strategy
The ramp and maximum spread determine both expected return and volatility. SCORE standardizes bankroll and risk while optimizing stakes within stated constraints. Importing a ramp from another game can invalidate the result.
2. Playing strategy
Rule-matched basic strategy and the exact deviation set are inputs. Changing count tags, true-count rounding or risk-averse indices may change expectation and variance. Execution errors are not automatically included unless the simulation models them.
3. Rules
Payment of 3:2 versus 6:5, S17 versus H17, DAS, surrender and restrictions on doubling or splitting affect the starting expectation. Dealer-blackjack settlement can also change strategy. A favorable rule does not compensate automatically for every unfavorable condition.
4. Penetration
Penetration is the portion of the shoe dealt before shuffling. Deeper dealing can create more observable variation and more opportunities to act at useful counts. Its value depends on the entire model, including the number of players and whether negative counts are played.
5. Risk
Formal SCORE uses a full-Kelly risk convention of approximately 13.5% ruin for its standardized bankroll. That is deliberately a comparison basis, not a declaration that this risk is acceptable. A person’s actual bankroll, constraints and chosen risk can be different, so the required stakes and practical outcome will differ.
A reproducible comparison example
Suppose two simulations use identical SCORE assumptions. Game A has SCORE 20 and Game B has SCORE 40. Under the standard 100-round rate and $10,000 benchmark, the second has twice the standardized expected win rate. It does not mean Game B wins twice as many sessions or guarantees profit.
QFIT also documents the relationship SCORE = DI² under the standardized framework. A DI of 5 corresponds to SCORE 25; a DI of 8 corresponds to SCORE 64. The squaring makes differences look larger, so compare the underlying assumptions before comparing the values.
If bankroll, risk, ramp constraints or observation policy differ, first normalize them or rerun the simulations. Otherwise the larger number may reflect different inputs rather than a better opportunity.
The standard assumptions and their limits
The $10,000 bankroll and approximately 13.5% ruin risk allow results from different games to share a basis. The usual presentation treats 100 rounds as an hourly comparison rate, so the number can be read as standardized dollars per 100 rounds under the model. Actual tables may run faster or slower, and actual stakes need not match the optimized schedule.
QFIT distinguishes c-SCORE, a comparative measure calculated without imposing the original bankroll and risk requirements, from formal SCORE. Authors should identify which measure they report rather than using the words interchangeably.
A SCORE result also does not certify that conditions are available, tolerated or executable. It summarizes modeled expectation and risk; it does not establish legality, permission, longevity or payment.
Using the comparison idea recreationally
A recreational player can compare payout, dealer rules, doubling, splitting and table minimums without calculating SCORE. That is ordinary game comparison. If the game remains negative expectation, a higher-quality rule set can reduce expected loss but does not create advantage-play SCORE.
Check the rules before playing
Record the exact payout, S17/H17, deck count, DAS, surrender, split limits and dealer-blackjack procedure. If information is missing, do not borrow a number from a table that merely looks similar.
Separate skill from table quality
Correct play cannot repair every poor rule, and good conditions cannot repair repeated execution errors. Evaluate both. Results from a few sessions do not show whether either factor is sound.
Keep actual money separate from the benchmark
The $10,000 figure defines the comparison; it is not a recommended playing bankroll. Use affordable limits for recreation. A long-run advantage model requires its own bankroll and risk analysis, including expenses and uncertainty about real conditions.
SCORE, house edge and c-SCORE
House edge or initial-bet advantage measures expectation relative to initial wagers under stated play. It does not include the volatility and betting ramp needed to compare counting approaches. SCORE adds a standardized risk-and-betting framework; c-SCORE provides a related comparison without the original bankroll and risk constraints.
Two games with similar initial-bet advantage can have different SCORE because penetration and betting opportunities change the optimized return and variance. Conversely, a high SCORE calculated with one spread cannot be compared blindly with a result using different constraints.
What to retain from the metric
Use SCORE only with a source, version and complete assumptions. Treat it as a model output, not a slogan. A higher value under matched inputs describes a better standardized return-risk tradeoff; it does not guarantee a positive finite result.
For direct calculations, use simulation output that states expected win, standard deviation, stakes, rules and penetration. Continue with the Desirability Index and risk of ruin.
SCORE and playing deviations
SCORE compares the full game model. The Illustrious 18 and Fab Four are compact sets of count-based deviations associated with Schlesinger’s work. They can be included as playing inputs, but learning them does not generate a SCORE number by itself.
Use indices that match the count, rules and conversion method. A risk-averse deviation can trade a small amount of expected value for lower variance in a specified betting context; it should not be mixed casually with an unrelated simulation.
Frequently asked questions
What does SCORE stand for?
Standardized Comparison of Risk and Expectation. It compares fully specified blackjack advantage-play models under common assumptions.
Is SCORE only for professional counters?
The formal calculation is designed for technical comparison of counting games. Anyone can understand it, but it does not turn recreational play into advantage play.
Can SCORE be used for online blackjack?
Only when a persistent card process and all required inputs can be modeled. Independently generated RNG rounds do not support traditional shoe counting.
How is it different from house edge?
House edge describes expected return under stated rules and play. SCORE also standardizes betting and risk to compare advantage-play opportunities.
Why does penetration matter?
It affects how much of a persistent shoe is observed and how often modeled betting and playing opportunities arise before the shuffle.
Must I calculate SCORE to play blackjack?
No. Basic strategy and ordinary rule comparison do not require SCORE. Use it only when comparing complete advantage-play simulations.
Is the $10,000 bankroll a recommendation?
No. It is a standardization assumption. Formal SCORE also uses an approximately 13.5% ruin-risk convention, which may be unacceptable for a person’s actual funds.
